Inspiration
Short-form video is dominating the internet, but creating high-quality content consistently is hard. I noticed that many “faceless” Instagram and TikTok pages rely on repetitive templates or low-quality automation. I wanted to build a system that lets anyone go from a raw idea to a cinematic, AI-generated video — without editing skills, cameras, or complex tools.
The goal was simple: turn an idea into a scroll-stopping video in minutes using modern generative AI.
What it does
Automated AI-Influencer is an end-to-end pipeline that transforms a simple idea into a short, high-quality video.
I enter a subject, style, and background. The app generates:
- A detailed image prompt
- A structured 8-second video script
- Multiple AI-generated images to choose from
- A final video generated using the selected image as reference
The workflow is interactive, allowing iteration at every step instead of one-shot generation.
How we built it
I built the project as a modular AI pipeline:
- Gemini 3 Flash Preview is used for prompt reasoning and orchestration.
- Imagen 4.0 generates high-quality images.
- Veo 3.1 generates videos using a selected image as visual grounding.
- Next.js (App Router) handles the frontend and API routes.
Each model has a clear responsibility, keeping the system controllable and debuggable.
Challenges we ran into
One major challenge was orchestrating multiple generative models while safely passing data between them. Handling base64 images, temporary storage, async video generation, rate limits, and safety filters required careful design.
Another challenge was ensuring Gemini 3 was used meaningfully as a reasoning layer rather than just a text generator.
Accomplishments that we're proud of
- Built a complete idea → image → video pipeline as a solo developer
- Successfully integrated Gemini 3, Imagen 4.0, and Veo 3.1
- Enabled image selection to guide video generation
- Designed a clean, iterative user experience
- Demonstrated practical multi-model AI orchestration
What we learned
I learned how powerful Gemini 3 is when used for reasoning and orchestration rather than simple text output. I also gained experience designing multi-stage generative systems with human-in-the-loop control.
This project reinforced the importance of modularity and transparency in AI-driven creative tools.
What's next for Automated Ai-Influencer
Next, I plan to:
- Add image references directly into video prompts
- Support custom starting and ending frames for smoother video flow
- Enable multi-scene video generation
- Add creator presets for platforms like Instagram Reels and YouTube Shorts
The long-term vision is to make Automated AI-Influencer a reliable creative assistant for building faceless content at scale.
Log in or sign up for Devpost to join the conversation.